Network ties and the success of open source software development

Highlights

• Co-membership among OSS projects is effective for building network ties.

• Network ties can channel knowledge and expertise and contribute to OSS success.

• Network ties have a diminishing return to OSS success.

• Leader–follower ties and follower–leader ties are more beneficial to OSS success.

• Network ties to projects of later development stages are more beneficial.

Abstract

Prior network-based research on open source software (OSS) development has focused on the benefit of network ties and assumed all network ties play the same role. We adopt a fine-grained view of network relations to investigate the impact of network ties on the success of OSS development. Through examining the development of OSS projects hosted by SourceForge, we find that co-membership among project teams is an effective mechanism for building network ties, through which knowledge and expertise flows across projects in OSS community and, therefore, contributes to the success of OSS development. However, network ties among projects not only confer benefit, but also incur various cost, and due to the different growth patterns of cost and benefit, network ties have a diminishing return to project success. In addition, we find network ties of leader–follower type and follower–leader type are more beneficial to OSS success than other types of ties, and network ties connecting to projects of later development stages are more beneficial than those connecting to projects of earlier stages. Our study provides useful guidelines and suggestions as to how to leverage the knowledge and expertise of others for successful development of OSS projects.

Introduction

Despite its deviation from the principles and practices advocated by traditional software development, open source software (OSS) development has gained increasing recognition (Raymond, 1999, Scacchi et al., 2006, Singh et al., 2011a). Over time, a significant amount of research has been conducted to investigate the success of OSS development from various perspectives (e.g., Colazo and Fang, 2010, Paulson et al., 2004, Raymond, 1999, Scacchi, 2004). In particular, realizing that social behavior and performance are not isolated but relational (Burt, 1992), researchers have adopted the social network approach to examine the success of OSS projects (Colazo, 2010, Grewal et al., 2006, Singh et al., 2011a).

For software development, knowledge and expertise are inherently rooted in the minds of individual members (Robert et al., 2008, Singh and Tan, 2011). Network ties can facilitate knowledge flow and exchange across OSS projects, and eventually influence the success of OSS projects. However, a closer examination of this literature reveals a theoretical gap: most studies focus on the benefit of network ties with few considering the cost of network ties; more importantly, although network ties have been found to affect OSS development in general, it is not clear which types of ties are more beneficial. We intend to bridge the above research gap in this research. By doing so, our study not only contributes to the network-based research on OSS success, but also bears practical implications as to how to leverage knowledge and experience of the others for successful development of OSS projects.

Specifically, we adopt a fine-grained view of network relations to examine the following three issues. First, we propose that project co-members establish network ties among OSS projects and co-membership is an effective mechanism for knowledge flow across OSS projects. Second, we argue that network ties confer benefits as well as incur obligations and costs, and this is particularly relevant for OSS development since network ties are built through project co-memberships. Due to different growth patterns of cost and benefit, OSS project success is not a linear but a concave function of the number of network ties, i.e., network ties have a diminishing return to the success of OSS projects. Third, realizing network ties differ in their capacity to channel and assimilate information and knowledge, we categorize network ties based on the roles played by the co-members as well as the development stages of connected projects, and further investigate which types of ties are more beneficial to OSS development.

The rest of the paper is organized as follows. In the next section, we introduce the background of this study and explicate the importance of co-memberships for knowledge flow across OSS projects. This is followed by Section 3 on theory and hypotheses development. Section 4 describes our dataset, variables, and methodology for hypotheses testing, and estimation results are presented in Section 5. In Section 6, we discuss contributions and limitations of the study, and offer directions for future research. The paper is concluded in Section 7.